말소리장애 아동 하위집단 간 음운 및 말산출 차이와 예측지표 분석: 예비연구
Phonological and speech production differences and predictive indicators among children with speech sound disorders: Preliminary study
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 연구는 언어지체 유무에 따라 분류된 말소리장애 세 하위집단 간 음운 및 말산출 지표의 차이를 분석하고, 집단을 변별하는 예측지표를 탐색하고자 하였다. 연구 대상은 만 3세에서 6세의 말소리장애 아동 48명(순수 말소리장애 22명, 표현언어지체 8명, 수용 및 표현언어지체 18명)이 참여하였다. 음운 및 말산출 특성을 분석하기 위해PMLU(Phonological Mean Length of Utterance), PWP(Proximity Whole Word Proximity), PWC(Proportion of Whole-Word Correctness), AMR(Alternate Motion Rate), SMR(Sequential Motion Rate), MPT(Maximum Phonation Time) 등의 지표를사용하였고, 통계분석은 R 프로그램(version 4.5.1)을 활용하였다. 집단 간 차이는 PERMANOVA(Permutational Multivariate Analysis of Variance)로 검정하였으며, 집단 구분을 위한 핵심 예측변수는 균형 랜덤 포레스트 모형으로분석하였다. PERMANOVA 결과, 통계적으로 유의한 다변량 차이는 나타나지 않았으나, 표현언어지체 집단은AMR, SMR, MPT 수행에서 다른 집단에 비해 낮은 평균을 보였다. 균형 랜덤 포레스트 분석에서는 PMLU, SMR, PWC, AMR이 주요한 예측 변수로 나타났다. 이러한 결과는 말소리장애의 평가 및 중재에 있어 단순 정확도 외에도음운 및 말산출 지표를 적극 활용한 다차원적 접근이 필요함을 시사한다.
Abstract
This study aimed to examine differences in phonological and speech production measures among three subgroups of children with speech sound disorders (SSD), categorized by language delay type: (1) pure SSD, (2) SSD with expressive language delay, and (3) SSD with both receptive and expressive language delay. This study also sought to identify the key predictors that distinguish these subgroups. Forty-eight children aged 3–6 years of age participated: 22 with pure SSD, 8 with expressive delay, and 18 with receptive–expressive delay. Measures included Phonological Mean Length of Utterance (PMLU), Proximity Whole Word Proximity (PWP), Proportion of Whole-Word Correctness (PWC), Alternate Motion Rate (AMR), Sequential Motion Rate (SMR), and Maximum Phonation Time (MPT) to assess phonological and speech motor abilities. Statistical analyses were conducted using R (version 4.5.1). Group differences were analyzed using Permutational Multivariate Analysis of Variance (PERMANOVA), and predictive variables were explored with a balanced random forest model. PERMANOVA revealed no statistically significant multivariate differences among the subgroups; however, children with expressive language delays demonstrated lower performance on the AMR, SMR, and MPT. Balanced random forest analysis identified PMLU, SMR, PWC, and AMR as the most important predictors of subgroup classification. These findings highlight the importance of using multidimensional measures—beyond simple accuracy—in evaluating and treating SSD. A comprehensive approach incorporating both phonological and speech motor features may enhance the precision of clinical decision-making and support individualized intervention planning.